Pub Date : 2026-04-01Epub Date: 2026-02-10DOI: 10.1016/j.biosystemseng.2026.104416
Yu Zhang, Fangming Tian, Feng Tan
This study addresses the challenge of predicting ground temperature in the active layer of seasonally frozen ground in Heilongjiang province, China. We optimised the ground temperature sensing network deployment scheme by analysing the temperature change characteristics of soil layers at different depths. On this basis, a ground temperature prediction model based on LSTM algorithm is proposed. The effective monitoring and prediction of the active layer of permafrost during spring field operations was achieved by optimising the number and location of probe deployments. The study successfully established an optimised four-sensor deployment scheme (0 m, 0.40 m, 0.75 m, and 3.40 m), which reduces deployment costs by 82% compared to the traditional setup. Experimental results show that the proposed LSTM model, based on this scheme, achieves high predictive accuracy (MAE = 0.0667, R2 = 0.9996). It indicates that the model performs well in predicting soil temperatures at different depths, which provides a scientific basis for agricultural cultivation and helps improve crop yields. It also offers technical support for seasonal frozen ground management and agricultural production.
{"title":"Optimisation of temperature sensing network deployment and prediction modelling of the active layer in Heilongjiang seasonally frozen ground during spring","authors":"Yu Zhang, Fangming Tian, Feng Tan","doi":"10.1016/j.biosystemseng.2026.104416","DOIUrl":"10.1016/j.biosystemseng.2026.104416","url":null,"abstract":"<div><div>This study addresses the challenge of predicting ground temperature in the active layer of seasonally frozen ground in Heilongjiang province, China. We optimised the ground temperature sensing network deployment scheme by analysing the temperature change characteristics of soil layers at different depths. On this basis, a ground temperature prediction model based on LSTM algorithm is proposed. The effective monitoring and prediction of the active layer of permafrost during spring field operations was achieved by optimising the number and location of probe deployments. The study successfully established an optimised four-sensor deployment scheme (0 m, 0.40 m, 0.75 m, and 3.40 m), which reduces deployment costs by 82% compared to the traditional setup. Experimental results show that the proposed LSTM model, based on this scheme, achieves high predictive accuracy (MAE = 0.0667, R<sup>2</sup> = 0.9996). It indicates that the model performs well in predicting soil temperatures at different depths, which provides a scientific basis for agricultural cultivation and helps improve crop yields. It also offers technical support for seasonal frozen ground management and agricultural production.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"264 ","pages":"Article 104416"},"PeriodicalIF":5.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146171759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-02-09DOI: 10.1016/j.biosystemseng.2025.104380
David Brunner , Marie Bordes , Elisabeth Mayrhuber , Stephan M. Winkler , Viktoria Dorfer , Maciej Oczak
Pose estimation is a popular computer vision method for the automated analysis of animals in observational studies. Since pose estimation is a challenging task, the use of pre-trained models is widespread. However, if the shift between the domains of pre-training and application is too large, fine-tuning of the models is necessary. Pose estimation is based on keypoints, the annotation of which is costly in terms of time and effort. In most realistic settings, the available resources for the annotation of video frames are limited. Therefore, it is crucial to maximise the utility of a small number of labelled frames. This study proposes skeleton integrity, a method for selecting the frames with the highest utility for fine-tuning a pose estimation model. It works by analysing the keypoint structure of the pre-trained model's predictions and only requires the initial preparation of a single labelled frame. The method was evaluated in the context of a study on social behaviour in pigs. Skeleton integrity was used to extract a selection of 100 high-utility frames (895 pig instances) from a large dataset recorded in the study. A detailed analysis was performed, showing that frame utility is determined by variations in keypoint visibility, crowding and the resolution of the pigs. An empirical study showed that a ViTPose model fine-tuned on a skeleton integrity-based selection outperformed the same model fine-tuned on a random selection by at least 2.51% in average precision and 3.48% in average recall, underlining the importance of targeted data selection for low-data fine-tuning.
{"title":"Skeleton integrity: A method for the efficient fine-tuning of pose estimation models for pigs","authors":"David Brunner , Marie Bordes , Elisabeth Mayrhuber , Stephan M. Winkler , Viktoria Dorfer , Maciej Oczak","doi":"10.1016/j.biosystemseng.2025.104380","DOIUrl":"10.1016/j.biosystemseng.2025.104380","url":null,"abstract":"<div><div>Pose estimation is a popular computer vision method for the automated analysis of animals in observational studies. Since pose estimation is a challenging task, the use of pre-trained models is widespread. However, if the shift between the domains of pre-training and application is too large, fine-tuning of the models is necessary. Pose estimation is based on keypoints, the annotation of which is costly in terms of time and effort. In most realistic settings, the available resources for the annotation of video frames are limited. Therefore, it is crucial to maximise the utility of a small number of labelled frames. This study proposes skeleton integrity, a method for selecting the frames with the highest utility for fine-tuning a pose estimation model. It works by analysing the keypoint structure of the pre-trained model's predictions and only requires the initial preparation of a single labelled frame. The method was evaluated in the context of a study on social behaviour in pigs. Skeleton integrity was used to extract a selection of 100 high-utility frames (895 pig instances) from a large dataset recorded in the study. A detailed analysis was performed, showing that frame utility is determined by variations in keypoint visibility, crowding and the resolution of the pigs. An empirical study showed that a ViTPose model fine-tuned on a skeleton integrity-based selection outperformed the same model fine-tuned on a random selection by at least 2.51% in average precision and 3.48% in average recall, underlining the importance of targeted data selection for low-data fine-tuning.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"264 ","pages":"Article 104380"},"PeriodicalIF":5.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146171777","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-30DOI: 10.1016/j.biosystemseng.2026.104398
Akash Ajagekar , Yu Jiang , Fengqi You
Computer vision and Internet of Things (IoT) technologies offer robust solutions for plant phenotyping, but traditional mainstream segmentation methods often fail in high-density plantings with overlapping foliage. This study introduces an integrated phenotyping system combining automated data capture and high-temporal RGB-D imaging using off-the-shelf hardware (Intel RealSense D435 and Raspberry Pi) to generate 3D point clouds of lettuce under controlled greenhouse conditions. While recent agricultural applications have shown limited success and required domain-specific adaptations, Segment Anything Model (SAM) and FastSAM were demonstrated to achieve exceptional zero-shot segmentation performance for individual lettuce plants in high-density arrangements without additional training. This capability effectively addresses the traditional challenges of species-specific parameter tuning and extensive training data requirements and fine-tuning. By mapping 2D segmentation masks to corresponding 3D point clouds, the system accurately extracted key phenotypic traits, namely plant height, length, and width, from which area and volume were subsequently estimated, showing strong correlations with manual measurements for Rex and Rouxai lettuce cultivars. This high-temporal, non-destructive monitoring provided unique insights into plant growth dynamics. The study highlights distinct growth patterns among these cultivars, underscoring the importance of tailored phenotyping approaches to optimise crop management strategies. By addressing the limitations of existing phenotyping methods, this work advances precision agriculture technologies, offering a cost-effective and efficient solution for monitoring dynamic crop growth with potential applications across various crops and growing conditions.
{"title":"Computer vision and IoT based plant phenotyping and growth monitoring with 3D point clouds","authors":"Akash Ajagekar , Yu Jiang , Fengqi You","doi":"10.1016/j.biosystemseng.2026.104398","DOIUrl":"10.1016/j.biosystemseng.2026.104398","url":null,"abstract":"<div><div>Computer vision and Internet of Things (IoT) technologies offer robust solutions for plant phenotyping, but traditional mainstream segmentation methods often fail in high-density plantings with overlapping foliage. This study introduces an integrated phenotyping system combining automated data capture and high-temporal RGB-D imaging using off-the-shelf hardware (Intel RealSense D435 and Raspberry Pi) to generate 3D point clouds of lettuce under controlled greenhouse conditions. While recent agricultural applications have shown limited success and required domain-specific adaptations, Segment Anything Model (SAM) and FastSAM were demonstrated to achieve exceptional zero-shot segmentation performance for individual lettuce plants in high-density arrangements without additional training. This capability effectively addresses the traditional challenges of species-specific parameter tuning and extensive training data requirements and fine-tuning. By mapping 2D segmentation masks to corresponding 3D point clouds, the system accurately extracted key phenotypic traits, namely plant height, length, and width, from which area and volume were subsequently estimated, showing strong correlations with manual measurements for Rex and Rouxai lettuce cultivars. This high-temporal, non-destructive monitoring provided unique insights into plant growth dynamics. The study highlights distinct growth patterns among these cultivars, underscoring the importance of tailored phenotyping approaches to optimise crop management strategies. By addressing the limitations of existing phenotyping methods, this work advances precision agriculture technologies, offering a cost-effective and efficient solution for monitoring dynamic crop growth with potential applications across various crops and growing conditions.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"264 ","pages":"Article 104398"},"PeriodicalIF":5.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071124","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-02-17DOI: 10.1016/j.biosystemseng.2026.104419
Xueke An , Nianrui Liu , Xiang Li , Zhaolei Yang , Fangyan Wang , Hassan H.A. Mostafa , Yuliang Yun
Agricultural production faces growing challenges due to labour shortages and the demand for environmentally sustainable techniques. Traditional blanket spraying methods, while effective in extensive agricultural fields, often result in excessive pesticide application and environmental hazards. For high-value crops like tomatoes, spot spraying offers a more precise and eco-friendlier alternative. This study proposes a novel Spot Spray Robot (SSR) framework that integrates disease detection, fuzzy logic–based task prioritisation, and path optimisation. A lightweight YOLOv12n model was developed for robust tomato leaf disease detection. Enhanced with a Multiscale Wavelet Pooling Transformer (MWPT) and refined by a Unified-IoU loss with dynamic weighting, the model achieved a precision of 0.953 and a recall of 0.932. Retrieved targets were aggregated and assessed by a fuzzy logic inference system, which considered path cost, disease density, spray deposition risk, and reachability to generate optimised execution sequences. Dual-nozzle hardware was further incorporated to improve adaptability by aligning the nozzle type with the target area. Experimental evaluations were conducted in both controlled indoor and greenhouse settings. Comparative results showed that the proposed method significantly reduced execution time and trajectory length compared with baseline strategies such as First-Come-First-Serve, Greedy Nearest Neighbour, and TSP-2opt. The system achieved smoother trajectories, reduced redundant manipulator motion, and improved spraying continuity. This study demonstrates the potential of integrating intelligent detection, fuzzy priority scheduling, and robotic execution for precision agriculture, providing a scalable and environmentally sustainable method for targeted pesticide application.
{"title":"Target sequence optimisation for precision spraying using fuzzy logic and priority evaluation","authors":"Xueke An , Nianrui Liu , Xiang Li , Zhaolei Yang , Fangyan Wang , Hassan H.A. Mostafa , Yuliang Yun","doi":"10.1016/j.biosystemseng.2026.104419","DOIUrl":"10.1016/j.biosystemseng.2026.104419","url":null,"abstract":"<div><div>Agricultural production faces growing challenges due to labour shortages and the demand for environmentally sustainable techniques. Traditional blanket spraying methods, while effective in extensive agricultural fields, often result in excessive pesticide application and environmental hazards. For high-value crops like tomatoes, spot spraying offers a more precise and eco-friendlier alternative. This study proposes a novel Spot Spray Robot (SSR) framework that integrates disease detection, fuzzy logic–based task prioritisation, and path optimisation. A lightweight YOLOv12n model was developed for robust tomato leaf disease detection. Enhanced with a Multiscale Wavelet Pooling Transformer (MWPT) and refined by a Unified-IoU loss with dynamic weighting, the model achieved a precision of 0.953 and a recall of 0.932. Retrieved targets were aggregated and assessed by a fuzzy logic inference system, which considered path cost, disease density, spray deposition risk, and reachability to generate optimised execution sequences. Dual-nozzle hardware was further incorporated to improve adaptability by aligning the nozzle type with the target area. Experimental evaluations were conducted in both controlled indoor and greenhouse settings. Comparative results showed that the proposed method significantly reduced execution time and trajectory length compared with baseline strategies such as First-Come-First-Serve, Greedy Nearest Neighbour, and TSP-2opt. The system achieved smoother trajectories, reduced redundant manipulator motion, and improved spraying continuity. This study demonstrates the potential of integrating intelligent detection, fuzzy priority scheduling, and robotic execution for precision agriculture, providing a scalable and environmentally sustainable method for targeted pesticide application.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"264 ","pages":"Article 104419"},"PeriodicalIF":5.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147385518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2026-01-01DOI: 10.1016/j.biosystemseng.2025.104381
Elisabeth Mayrhuber , Kristina Maschat , David Brunner , Stephan M. Winkler , Maciej Oczak
Predicting the onset of farrowing in sows is critical for improving animal welfare and optimising farm management. Methods driven by explainable artificial intelligence for detecting nest-building behaviour and predicting time to farrowing using accelerometer data from ear tags are presented. These methods are evaluated on a dataset containing farm management data and accelerometer data of 179 sows. During data collection the animals were kept in three different pen types with the possibility of temporary crating. By combining acceleration metrics with prepartum examinations and farm management data, a two-stage model was developed that first detects the onset of nest-building and subsequently predicted the remaining time until farrowing. Various methods, including cumulative sum (CUSUM), Bayesian estimation of abrupt change, seasonality, and trend (BEAST), and a custom model (NestDetect), were compared for nest-building detection, while symbolic regression and deep learning were used to predict farrowing time. For 82.6 % of the sows, it was possible to detect the start of nest-building behaviour in a 48-h window before the onset of farrowing. When nest-building was detected correctly, symbolic regression was able to predict the remaining time to farrowing with a mean absolute error of 9.4 h and delivered interpretable results, while NNs achieved a mean absolute error of 9.6 h without being inherently interpretable. This work emphasises the importance of model interpretability and explainability in precision livestock farming, highlighting that transparent models can facilitate timely, data-driven interventions, while having the same prediction power as non-interpretable models.
{"title":"Improved and interpretable accelerometer-based farrowing prediction","authors":"Elisabeth Mayrhuber , Kristina Maschat , David Brunner , Stephan M. Winkler , Maciej Oczak","doi":"10.1016/j.biosystemseng.2025.104381","DOIUrl":"10.1016/j.biosystemseng.2025.104381","url":null,"abstract":"<div><div>Predicting the onset of farrowing in sows is critical for improving animal welfare and optimising farm management. Methods driven by explainable artificial intelligence for detecting nest-building behaviour and predicting time to farrowing using accelerometer data from ear tags are presented. These methods are evaluated on a dataset containing farm management data and accelerometer data of 179 sows. During data collection the animals were kept in three different pen types with the possibility of temporary crating. By combining acceleration metrics with prepartum examinations and farm management data, a two-stage model was developed that first detects the onset of nest-building and subsequently predicted the remaining time until farrowing. Various methods, including cumulative sum (CUSUM), Bayesian estimation of abrupt change, seasonality, and trend (BEAST), and a custom model (NestDetect), were compared for nest-building detection, while symbolic regression and deep learning were used to predict farrowing time. For 82.6 % of the sows, it was possible to detect the start of nest-building behaviour in a 48-h window before the onset of farrowing. When nest-building was detected correctly, symbolic regression was able to predict the remaining time to farrowing with a mean absolute error of 9.4 h and delivered interpretable results, while NNs achieved a mean absolute error of 9.6 h without being inherently interpretable. This work emphasises the importance of model interpretability and explainability in precision livestock farming, highlighting that transparent models can facilitate timely, data-driven interventions, while having the same prediction power as non-interpretable models.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"263 ","pages":"Article 104381"},"PeriodicalIF":5.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145882441","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2026-01-14DOI: 10.1016/j.biosystemseng.2026.104383
Zezhong Chen , Qiumei Yang , Deqin Xiao , Jiyan Wu , Manting Wu , Qiwei Hong
Detecting abnormal behaviours in pigs is crucial for enhancing pig welfare. Current research on pig anomaly detection primarily relies on supervised learning methods, facing challenges such as limited generalisability, the complexity of sample annotation, and the inability to cover all abnormal scenarios. To tackle these challenges, an unsupervised video anomaly detection algorithm for pigs based on future frame prediction (PigVADNet) is proposed. PigVADNet is developed to address the unpredictability of abnormalities in pig production. It accurately predicts normal pig behaviours by learning from video frames depicting normal pig behaviours. When the video frames capture abnormal behaviours, there is a significant increase in prediction error, which enables the detection of anomalies in pigs. The model employs a generative adversarial network architecture consisting of a pig image generator, discriminator, and motion information extraction module. The generator leverages a U-Net with an SSPCAB (Spatial and Spectral Pyramid Channel Attention Block) module to predict future frames. The discriminator improves the generator via adversarial learning, ensuring realistic frame generation. The motion extraction module, combined with appearance and motion consistency losses, enhances the prediction of appearance and motion. Finally, the difference between predicted and real frames is evaluated to detect pig abnormalities. The model achieved an AUC (Area Under the ROC Curve) of 95.1 % on the Pig Video Anomaly Detection Dataset. The experimental results demonstrate that this approach can automatically detect pig anomalies without relying on labelled data. It enables timely interventions to enhance pig welfare and optimise production efficiency.
{"title":"A novel unsupervised algorithm for pig anomaly detection using video frame prediction","authors":"Zezhong Chen , Qiumei Yang , Deqin Xiao , Jiyan Wu , Manting Wu , Qiwei Hong","doi":"10.1016/j.biosystemseng.2026.104383","DOIUrl":"10.1016/j.biosystemseng.2026.104383","url":null,"abstract":"<div><div>Detecting abnormal behaviours in pigs is crucial for enhancing pig welfare. Current research on pig anomaly detection primarily relies on supervised learning methods, facing challenges such as limited generalisability, the complexity of sample annotation, and the inability to cover all abnormal scenarios. To tackle these challenges, an unsupervised video anomaly detection algorithm for pigs based on future frame prediction (PigVADNet) is proposed. PigVADNet is developed to address the unpredictability of abnormalities in pig production. It accurately predicts normal pig behaviours by learning from video frames depicting normal pig behaviours. When the video frames capture abnormal behaviours, there is a significant increase in prediction error, which enables the detection of anomalies in pigs. The model employs a generative adversarial network architecture consisting of a pig image generator, discriminator, and motion information extraction module. The generator leverages a U-Net with an SSPCAB (Spatial and Spectral Pyramid Channel Attention Block) module to predict future frames. The discriminator improves the generator via adversarial learning, ensuring realistic frame generation. The motion extraction module, combined with appearance and motion consistency losses, enhances the prediction of appearance and motion. Finally, the difference between predicted and real frames is evaluated to detect pig abnormalities. The model achieved an AUC (Area Under the ROC Curve) of 95.1 % on the Pig Video Anomaly Detection Dataset. The experimental results demonstrate that this approach can automatically detect pig anomalies without relying on labelled data. It enables timely interventions to enhance pig welfare and optimise production efficiency.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"263 ","pages":"Article 104383"},"PeriodicalIF":5.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Electrostatic spraying technology enhances spray efficiency and reduces airborne drift by imparting electrical charges to droplets, which increases their attraction to crop canopies. However, determining the best configuration for an electrostatic pesticide sprayer is difficult as numerous parameters impact spray efficiency. Experimental optimisation is resource-intensive and time-consuming, which makes computational modelling an excellent alternative optimisation method. This study developed a computational fluid dynamics (CFD) model to predict droplet charge-to-mass ratio (CMR), canopy deposition, and downwind drift for electrically charged sprays. The model was validated against canopy deposition and airborne drift measurement data collected in a wind tunnel at wind speeds of 0 and 2.24 m s−1 using five hollow-cone nozzles and a 50 mm diameter electrode held at an applied voltage of 20 kV DC. Results showed that the model could predict the average canopy deposition from an electrostatic spraying system at specific locations within the canopy with average relative errors of 40.3 % and 58.8 % at wind speeds of 0 and 2.24 m s−1, respectively. At a wind speed of 2.24 m s−1, the model acceptably predicted airborne drift deposits up to a 0.70 m height, with an average relative error of 50.1 % for the validated cases; however, prediction errors increased substantially above this height. These findings demonstrate that CFD modelling is a promising method for optimising electrostatic spraying system configurations to maximise spray efficiency and minimise airborne drift, especially in low-wind environments, such as greenhouses.
静电喷雾技术提高了喷雾效率，并通过向液滴传递电荷来减少空气漂移，从而增加了液滴对作物冠层的吸引力。然而，确定静电农药喷雾器的最佳配置是困难的，因为许多参数影响喷雾效率。实验优化是一种资源密集和耗时的优化方法，这使得计算建模成为一种很好的替代优化方法。本研究开发了一个计算流体动力学（CFD）模型来预测带电喷雾的电荷质量比（CMR）、冠层沉积和顺风漂移。在风速为0和2.24 m s - 1的风洞中，使用5个空心锥喷嘴和直径为50 mm的电极，在20 kV直流电压下，对该模型进行了验证。结果表明：在风速为0 m s−1和2.24 m s−1的条件下，该模型能够预测林冠内特定位置静电喷涂系统的平均林冠沉降，平均相对误差分别为40.3%和58.8%。在风速为2.24 m s−1时，该模型可接受地预测高达0.70 m的空中漂移沉积物，验证案例的平均相对误差为50.1%；然而，在此高度以上，预测误差大大增加。这些研究结果表明，CFD建模是一种很有前途的方法，可以优化静电喷涂系统配置，以最大限度地提高喷雾效率，减少空气漂移，特别是在低风环境中，如温室。
{"title":"CFD modelling of an electrostatic spraying system to optimise pesticide spray efficiency and reduce drift","authors":"Matthew Herkins , Lingying Zhao , Heping Zhu , Hongyoung Jeon","doi":"10.1016/j.biosystemseng.2025.104378","DOIUrl":"10.1016/j.biosystemseng.2025.104378","url":null,"abstract":"<div><div>Electrostatic spraying technology enhances spray efficiency and reduces airborne drift by imparting electrical charges to droplets, which increases their attraction to crop canopies. However, determining the best configuration for an electrostatic pesticide sprayer is difficult as numerous parameters impact spray efficiency. Experimental optimisation is resource-intensive and time-consuming, which makes computational modelling an excellent alternative optimisation method. This study developed a computational fluid dynamics (CFD) model to predict droplet charge-to-mass ratio (CMR), canopy deposition, and downwind drift for electrically charged sprays. The model was validated against canopy deposition and airborne drift measurement data collected in a wind tunnel at wind speeds of 0 and 2.24 m s<sup>−1</sup> using five hollow-cone nozzles and a 50 mm diameter electrode held at an applied voltage of 20 kV DC. Results showed that the model could predict the average canopy deposition from an electrostatic spraying system at specific locations within the canopy with average relative errors of 40.3 % and 58.8 % at wind speeds of 0 and 2.24 m s<sup>−1</sup>, respectively. At a wind speed of 2.24 m s<sup>−1</sup>, the model acceptably predicted airborne drift deposits up to a 0.70 m height, with an average relative error of 50.1 % for the validated cases; however, prediction errors increased substantially above this height. These findings demonstrate that CFD modelling is a promising method for optimising electrostatic spraying system configurations to maximise spray efficiency and minimise airborne drift, especially in low-wind environments, such as greenhouses.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"263 ","pages":"Article 104378"},"PeriodicalIF":5.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145882464","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Extracting poplar seed morphological phenotypes is a core task in modern poplar breeding research. Accurate seed image segmentation is crucial for phenotype extraction and data quality. However, the small size of poplar seeds and their tendency to form dense clusters challenge the accuracy of current segmentation methods. Unlike current approaches that struggle with small-target segmentation and boundary delineation, this study develops the MP-Seed semantic segmentation algorithm, which combines a small-target attention module (based on Layer Across Feature Map Attention) with a multi-task learning mechanism that integrates boundary features. This novel integration targets small-seed key regions, fuses boundary features, and refines predictions to precisely segment densely clustered seeds, achieving superior accuracy and fine-grained delineation compared to current single-task methods. To address low efficiency and accuracy in poplar seed morphological phenotype extraction, this study further proposes a high-throughput extraction method leveraging the MP-Seed algorithm. To analyse the phenotypic data, an SVM classification model classifies eight types of poplar seeds. Experimental validation shows that the MP-Seed algorithm outperforms current methods on the test set, achieving Seed_IoU of 94.1 %, mIoU of 97.2 %, and Reference_IoU of 97.6 %. The high-throughput phenotyping method measures seed length and width with relative errors within 2.72 % versus manual measurements and extracts ten morphological traits at about 18.3 seeds per second. The overall classification accuracy reaches 91.1 %. Overall, this study provides technical support for accurate poplar seed segmentation and efficient morphological phenotype extraction, offering a valuable reference for other seed morphological phenotype research and analysis.
{"title":"From segmentation to classification: Morphological phenotype extraction and classification analysis of tiny poplar seeds using the MP-Seed segmentation algorithm","authors":"Zanpeng Li, Mengmeng Qiao, Xiwei Wang, Mubikayi Muhong Horly, Maocheng Zhao, Bin Wu","doi":"10.1016/j.biosystemseng.2025.104376","DOIUrl":"10.1016/j.biosystemseng.2025.104376","url":null,"abstract":"<div><div>Extracting poplar seed morphological phenotypes is a core task in modern poplar breeding research. Accurate seed image segmentation is crucial for phenotype extraction and data quality. However, the small size of poplar seeds and their tendency to form dense clusters challenge the accuracy of current segmentation methods. Unlike current approaches that struggle with small-target segmentation and boundary delineation, this study develops the MP-Seed semantic segmentation algorithm, which combines a small-target attention module (based on Layer Across Feature Map Attention) with a multi-task learning mechanism that integrates boundary features. This novel integration targets small-seed key regions, fuses boundary features, and refines predictions to precisely segment densely clustered seeds, achieving superior accuracy and fine-grained delineation compared to current single-task methods. To address low efficiency and accuracy in poplar seed morphological phenotype extraction, this study further proposes a high-throughput extraction method leveraging the MP-Seed algorithm. To analyse the phenotypic data, an SVM classification model classifies eight types of poplar seeds. Experimental validation shows that the MP-Seed algorithm outperforms current methods on the test set, achieving Seed_IoU of 94.1 %, mIoU of 97.2 %, and Reference_IoU of 97.6 %. The high-throughput phenotyping method measures seed length and width with relative errors within 2.72 % versus manual measurements and extracts ten morphological traits at about 18.3 seeds per second. The overall classification accuracy reaches 91.1 %. Overall, this study provides technical support for accurate poplar seed segmentation and efficient morphological phenotype extraction, offering a valuable reference for other seed morphological phenotype research and analysis.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"263 ","pages":"Article 104376"},"PeriodicalIF":5.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145882440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2026-01-19DOI: 10.1016/j.biosystemseng.2025.104358
Wensheng Sun , Shujuan Yi , Hailong Qi , Yifei Li , Zhibo Dai , Yupeng Zhang , Song Wang , Yunxiao Liu
To solve the problem of large seeding volume of the planter under the dense planting mode of maize, and the high requirements on the seed-supplying capacity of the seed-supplying device during high-speed operation, an air-assisted spiral seed-supply device is designed. The combination of spiral seed relocation and airflow seed delivery is used for efficient seed supply. The coupling of discrete element method and computational fluid dynamics (DEM-CFD) was used to investigate the influence of different pipe lengths, sleeve axial openings, and spiral shaft guides on the device's ability to seed-supply. A quadratic regression model was fitted between pipe lengths, sleeve axial openings, and spiral shaft guides, and seed-supply performance indexes, to obtain the optimal parameter combinations of the device. The effects of different types of seeds and the rotational speed of the spiral shaft on the performance of the device were investigated through bench tests. The results show that the optimal combination of structural parameters of the device is 44.153 mm of sleeve axial opening, 56.228 mm of spiral shaft guide, and 644.998 mm of pipe length, and the seed supply rate, the coefficient of variation of seed supply rate stability, and the seed breakage rate under the simulation validation are 24.72 g s−1, 2.04 %, and 1.66 % respectively, and the deviation of the bench validation results from the simulation validation results is 0.15 g s−1, 0.08 %, and 0.2 % respectively, which verifies the validity of the optimisation results of the simulation test parameters.
为解决玉米密集种植模式下播种机播种量大，高速运行时对供种装置供种能力要求高的问题，设计了一种气助式螺旋供种装置。采用螺旋送种和气流送种相结合的方式，实现高效送种。采用离散元法和计算流体力学（DEM-CFD）相结合的方法，研究了不同管道长度、套筒轴向开口和螺旋轴导轨对装置供种能力的影响。通过对管道长度、套筒轴向开口、螺旋轴导轨与供种性能指标进行二次回归模型拟合，得到该装置的最优参数组合。通过台架试验，研究了不同种子类型和螺旋轴转速对装置性能的影响。结果表明,结构参数的最佳组合套管轴向开放的设备是44.153毫米,56.228毫米的螺旋轴向导,和644.998毫米的管道长度,和种子供应率、种子供应率的变异系数稳定,和种子破碎率仿真验证以下24.72 g s−1 2.04%和1.66%分别和替补席上的偏差验证仿真结果验证结果0.15 g s−1)0.08%,和0.2%，验证了仿真试验参数优化结果的有效性。
{"title":"Design and experiment of air-assisted spiral seed-supply device for high-speed narrow-row dense planting of maize","authors":"Wensheng Sun , Shujuan Yi , Hailong Qi , Yifei Li , Zhibo Dai , Yupeng Zhang , Song Wang , Yunxiao Liu","doi":"10.1016/j.biosystemseng.2025.104358","DOIUrl":"10.1016/j.biosystemseng.2025.104358","url":null,"abstract":"<div><div>To solve the problem of large seeding volume of the planter under the dense planting mode of maize, and the high requirements on the seed-supplying capacity of the seed-supplying device during high-speed operation, an air-assisted spiral seed-supply device is designed. The combination of spiral seed relocation and airflow seed delivery is used for efficient seed supply. The coupling of discrete element method and computational fluid dynamics (DEM-CFD) was used to investigate the influence of different pipe lengths, sleeve axial openings, and spiral shaft guides on the device's ability to seed-supply. A quadratic regression model was fitted between pipe lengths, sleeve axial openings, and spiral shaft guides, and seed-supply performance indexes, to obtain the optimal parameter combinations of the device. The effects of different types of seeds and the rotational speed of the spiral shaft on the performance of the device were investigated through bench tests. The results show that the optimal combination of structural parameters of the device is 44.153 mm of sleeve axial opening, 56.228 mm of spiral shaft guide, and 644.998 mm of pipe length, and the seed supply rate, the coefficient of variation of seed supply rate stability, and the seed breakage rate under the simulation validation are 24.72 g s<sup>−1</sup>, 2.04 %, and 1.66 % respectively, and the deviation of the bench validation results from the simulation validation results is 0.15 g s<sup>−1</sup>, 0.08 %, and 0.2 % respectively, which verifies the validity of the optimisation results of the simulation test parameters.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"263 ","pages":"Article 104358"},"PeriodicalIF":5.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023375","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2026-01-22DOI: 10.1016/j.biosystemseng.2026.104399
Jingang Han , Guobin Wang , Xinyu Xue , Cancan Song , Yubin Lan
As an emerging precision agriculture technology, UAV fertiliser application technology has been rapidly developed in recent years. However, existing UAV-based fertiliser applicators (UFAs) lack differentiated variable performance design for different discharge port outlets. To address this limitation, this study designed an UAV fertiliser applicator with adjustable fertiliser discharge. This UFA mainly consists of a flow-regulating fan and an adjustment module. The flow-regulating fan is installed at the bottom of the fertiliser tank to adjust fertiliser discharge rate. The regulating unit is installed at the bottom of the flow-regulating fan to adjust the differentiated fertiliser amount at the outlets. The motion model of fertiliser particles was established based on DEM, and used to analyse the influence of parameters such as the feeding angle, flow-regulating fan angle, and outlet angle on the effect pattern on the variation of fertiliser application rate at different outlets. Bench tests were conducted to verify the overall discharge performance and the differences among various outlets under different discharge rates and combinations of the regulating units. Simulation results showed that when the feeding angle is 70°, the flow-regulating fan angle is 15°, and the outlet angle is 15°, the coefficient of variation (CV) of the five outlets is 66.63 %, demonstrating that the UFA can achieve significant differentiation in fertiliser discharge among outlets. Bench tests showed that the average proportion of fertiliser discharged from individual outlets ranged from approximately 8.59 %–61.38 %, confirming substantial variability. This study can provide a reference for the research on variable UFAs with different outlets to change the amount of fertiliser applied.
{"title":"Design and optimisation of differentiated UAV-based fertiliser applicator","authors":"Jingang Han , Guobin Wang , Xinyu Xue , Cancan Song , Yubin Lan","doi":"10.1016/j.biosystemseng.2026.104399","DOIUrl":"10.1016/j.biosystemseng.2026.104399","url":null,"abstract":"<div><div>As an emerging precision agriculture technology, UAV fertiliser application technology has been rapidly developed in recent years. However, existing UAV-based fertiliser applicators (UFAs) lack differentiated variable performance design for different discharge port outlets. To address this limitation, this study designed an UAV fertiliser applicator with adjustable fertiliser discharge. This UFA mainly consists of a flow-regulating fan and an adjustment module. The flow-regulating fan is installed at the bottom of the fertiliser tank to adjust fertiliser discharge rate. The regulating unit is installed at the bottom of the flow-regulating fan to adjust the differentiated fertiliser amount at the outlets. The motion model of fertiliser particles was established based on DEM, and used to analyse the influence of parameters such as the feeding angle, flow-regulating fan angle, and outlet angle on the effect pattern on the variation of fertiliser application rate at different outlets. Bench tests were conducted to verify the overall discharge performance and the differences among various outlets under different discharge rates and combinations of the regulating units. Simulation results showed that when the feeding angle is 70°, the flow-regulating fan angle is 15°, and the outlet angle is 15°, the coefficient of variation (CV) of the five outlets is 66.63 %, demonstrating that the UFA can achieve significant differentiation in fertiliser discharge among outlets. Bench tests showed that the average proportion of fertiliser discharged from individual outlets ranged from approximately 8.59 %–61.38 %, confirming substantial variability. This study can provide a reference for the research on variable UFAs with different outlets to change the amount of fertiliser applied.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"263 ","pages":"Article 104399"},"PeriodicalIF":5.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022869","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号